Can education research be scientific?
Is it possible to apply scientific methods to education research or is any such attempt misguided?
One of the arguments in Gert Biesta’s book, The Beautiful Risk of Education, is that education is not certain. If we intervene in a particular way in the education of an individual student, we cannot be certain of the result. Instead, we need to take a more nuanced and individualised approach.
Such arguments populate the education debate more broadly. For instance there are those who point out that they or their child learnt to read without systematic phonics instruction. This is then presented as a falsification of the idea that we need schools to adopt systematic phonics instruction. Instead, we need to take a more individualised approach.
Often, protagonists will go further. A human sphere such as education is highly unpredictable, they claim, and so the methods used in the physical sciences, where hydrochloric acid added to calcium carbonate will reliably produce carbon dioxide, calcium chloride and water, are completely unsuitable.
These people do not understand science and they do not understand education research.
There is a state between being able to fully predict what will happen to every constituent part of a system and throwing your hands up and lamenting that nothing can be predicted. This is the realm of probability and statistical inference. It is what distinguishes inductive science from deductive maths. It is baked into the scientific method.
For instance, in reversible chemical reactions, it is impossible to predict what will happen to an individual molecule. The best we can do is produce some bulk statistics such as what happens on average or what proportion of molecules react. This is the stuff of basic chemistry.
But what of that most fundamental of the sciences, physics? Surely, physics is completely deterministic?
This is where my undergraduate physics training comes in. In one extended investigation, a radioactive substance emitted gamma particles. We were asked to determine whether the pattern of emissions matched the one predicted by theory. No single emission was predictable but, over time, the sum of these emissions made a predictable pattern. I remember using an early version of Excel to test whether our data matched this pattern and to what degree of uncertainty.
For an astronomy literature review, we were asked to decide whether one type of active galaxy was the same as a wider class of galaxy only with a specific orientation towards us. Again, this is a question of statistics and probability.
I was struck by this when reading an item in the New Scientist. Provided the signal cannot be attributed to noise, researchers may have discovered an early gamma ray burst by monitoring a faraway, and therefore very old, galaxy. However, if they did, it may indicate gamma ray bursts were common in the early universe because you would only expect a few such burst from a single galaxy in a million years. Alternatively, they may have been very lucky.
Over time, more data will sort through the options. The point is that science deals with uncertainty all the time. It is in its DNA. As an undergraduate physicist, the first thing I was taught was how to estimate the uncertainty in any measurements we made.
How does this relate to education? It means we should not despair at uncertainty. I don’t believe anyone has ever claimed that simply adopting systematic phonics will cause all children to read or that children cannot read by any other method. Instead, the claim is that the greatest proportion of children will learn to read this way and fewer will read through alternative methods. These are scientifically testable claims. Moreover, if there is a subgroup who respond better to an alternative - which I personally doubt - this is a scientifically testable question too. We can use the scientific method to establish this group exists and develop screens to see who is a member of this group. We can then tailor our teaching methods based on more than a hunch.
When people claim that science is a fundamentally inappropriate method for education research because humans are messy and complex and science does not account for that, there are two possibilities. The first is a stunning ignorance of what science actually is - an ignorance that has not progressed much further than stereotypes and children’s TV. The second is a desire to drive the incisive analysis of science out of town in case it upsets comfortable rituals and hierarchies. The reality is probably a bit of both.
Does this mean science is always done well in education? Certainly not. Most attempts to apply science to education are shockingly bad because they violate basic scientific principles such as varying only one factor at a time. Nevertheless, this does not prove it is a fruitless pursuit that we can label ‘positivism’ and throw in the bin. All this proves is that we need to do better. Much better.